The impact of observational weighting on the assimilation of GPS/MET bending angle

The sensitivity of Global Positioning System (GPS)/Meteorology (GPS/MET) bending angle assimilation to the specification of observational weightings is studied. Three weighings are used. The first is estimated based on background and observation differences. The second comes from a theoretical estimation of the errors in the calculation of radio occultation (RO) bending angles [Zou et al, 2002]. The third is given by Palmer et al. [2000]. A total of 52 single GPS/MET occultation experiments are carried out for each of these weightings. The relationships between the adjustment of the refractivity and the adjustments of the temperature, specific humidity, and surface pressure have been examined. Collocated radiosonde observations are used as independent data sources for the numerical evaluation of GPS/MET data assimilation results. The analysis increments of the specific humidity and the surface pressure have the same sign as that of the refractivity. The larger the analysis increments of the refractivity, the larger the analysis increments of the specific humidity and the surface pressure. Such a one-to-one correspondence is not found between the analysis increment of the temperature and the increment of the refractivity. The GPS/ MET bending angle assimilation with these three weightings all provide reasonable improvements to the temperature and specific humidity analyses compared with radiosonde data. The National Centers for Environmental Prediction (NCEP) background fields are wetter than radiosonde measurements between 800 and 450 hPa. This positive bias of the background specific humidity is largely removed after the GPS/MET data assimilation with all three weightings. However, it is found that GPS/MET bending angle assimilation slightly increased the difference between the surface pressure analysis and the radiosonde observations. Different analysis variables have different sensitivities to the weightings. Among them, the sensitivities of the temperature and the surface pressure are most significant. In a further analysis, the sensitivities of the data assimilation to the variation of the gravity and the vertical resolution of the assimilated data are presented. The sensitivity to the gravity is not very significant. However, the increase of the vertical resolution of the assimilated bending angle observations produces more accurate refractivity and temperature analyses, with little impact on surface pressure fields. Moreover, the inconsistency between two independent observations may cause the background fields to be positively biased from one type of observation but negatively biased from the other type of observation. Such an inconsistency has been seen to affect the evaluation of the GPS/MET bending angle assimilation results using radiosonde observations.